Ph.D. thesis by Kurt Højlund. The Diabetes Research Centre Department of Medical Endocrinology Odense University Hospital

Transcription

1 $ Skeletal Muscle Insulin Resistance in Type 2 Diabetes Investigated by Two Approaches: Studies of Insulin Signaling into Glycogen Synthase and Proteome Analysis Ph.D. thesis by Kurt Højlund The Diabetes Research Centre Department of Medical Endocrinology Odense University Hospital Faculty of Health Sciences University of Southern Denmark

2 PREFACE The PhD dissertation is based on studies performed at the Diabetes Research Centre, Department of Endocrinology, Odense University Hospital, the Centre for Proteome Analysis (CPA), University of Southern Denmark and the Copenhagen Muscle Research Centre (CMRC), Institute of Exercise and Sport Sciences, Department of Human Physiology, University of Copenhagen, Denmark during During these years the help and support from a number of coworkers and laboratory technicians has been a prerequisite for the completion of this work, and I owe them all many thanks. First of all I want to express my profound gratitude to my principal supervisor, professor Henning Beck-Nielsen, who introduced me to the field of diabetes research, and encouraged me to go into laboratory work. His extensive knowledge and great experience within the field of type 2 diabetes and his personal contact to numerous research groups all over the world has been of fundamental importance. I am very grateful for his never failing support, inspiring discussions and always positive and optimistic attitude. I am greatly indebted to my supervisor Jørgen F. P. Wojtaszewski for the opportunity to study the intracellular actions of insulin. Without his assistance with the planning and performance of these studies the completion of my work would not have been possible. I am grateful for his always competent advice and constructive and clever critiscism, and for being virtually on-line accessible whenever I need help or to discuss data or ideas. I want to express my sincere thanks to my supervisors Peter Mose Larsen and Stephen J. Fey for introducing me to the world of proteomics and allowing me to work in their laboratory. Their never-failing enthusiasm, friendly attitude and great hospitality has been a great support for me during critical phases in my experience with this novel tool for the study of type 2 diabetes. I wish to thank my good collegues and friends Krzysztof Wrzesinski, Peter Stæhr and Klaus Levin for their contributions to these and other studies, for always being ready to help, and for many hours of interesting discussions and a lot of fun. I also want to thank Michael Gaster, Klaus Bruusgaard, Bo Falck Hansen, professor Erik A. Richter, Marianne Poulsen, D Grahame Hardie, Kevin A Green and Kirsty Mustard for good discussions and their good collaboration and scientific contributions to my work. Aase Handberg, Flemming Dela, Jørgen Vinten, Christine Reynet, Jim McCormack and professor Peter Roepstorff are thanked for many helpful discussions and for their contributions to the proteomic study. Warm thanks to Jan Erik Henriksen and Ole Hother-Nielsen for their valuable advice in scientific and practical problems, and to other members of the Diabetes Research Group for their interest and 2

3 input in many discussions. I would like to gratefully acknowledge the extensive and skilful technical assistance of Lone Hansen, Charlotte B. Olsen, Karin Dyrgaard and Henny Hansen. I am very grateful for the excellent secretary assistance and support provided by Tine Christensen, Eva Lund, Janne Dyhr and Elsebeth Byrge. Furthermore, the staff at CPA are acknowledged for expert technical assistance, secretary help and for their hospitality whenever I visit CPA. Bettina Bolmgren, Jesper Birk and other persons at CMRC are acknowledged for excellent technical assistance. Warm thanks to the diabetic patients and control subjects who have been involved in these studies. Their patience and positive attitude toward our work is greatly appreciated. Financial support for the studies was provided by grants from the Clinical Research Institute (University of Southern Denmark), the Danish National Research Foundation, the Danish Diabetes Association, the Novo Nordisk Foundation, the Faculty of Health Sciences (University of Southern Denmark), Fonden til Lægevidenskabens Fremme, Grosserer M.Brogaard og Hustrus Mindefond, Eva og Hans Carl Adolf Holms Mindelegat and the Eli Lilly Foundation. Most of all, I wish to express my gratefulness to my wife, Henriette and my children Axel Valdemar and Astrid Emilie for their never ending love, support and patience throughout the creation of this thesis. 3

4 This Ph.D.-thesis is based on the following four papers, of which three have been published and one has given rise to an international patent application. I. Højlund K, Stæhr P, Hansen BF, Green KA, Hardie DG, Ricther EA, Beck-Nielsen H. and Wojtazsewski JFP. Increased phosphorylation of skeletal muscle glycogen synthase at NH 2 - terminal sites during physiological hyperinsulinemia in type 2 diabetes. Diabetes 2003;52: II. Højlund K, Mustard KJ, Stæhr P, Hardie DG, Beck-Nielsen H, Richter EA and Wojtaszewski JFP. AMPK activity and isoform protein expression are similar in muscle of obese subjects with and without type 2 diabetes. Am. J. Physiol. Endocrinol. Metab. In press (Oct 7, 2003). III. Højlund K, Poulsen M, Stæhr P, Bruusgaard K and Beck-Nielsen H. Effect of insulin on protein phosphatase 2A expression in muscle in type 2 diabetes. Eur. J. Clin. Invest. 2002;32: IV. Højlund K, Wrzesinski K, Mose Larsen P, Fey SJ, Roepstorff P, Handberg A, Dela F, Vinten J, McCormack JG, Reynet C and Beck-Nielsen H. Proteome analysis reveals phosphorylation ATP synthase beta-subunit in human skeletal muscle and proteins with potential roles in type 2 diabetes. J. Biol. Chem. 2003;278: Patent: Højlund K, Mose Larsen P, Fey SJ, Beck-Nielsen H and Wrzesinski K. Proteins in Type 2 Diabetes. WO 03/ A2. 4

5 Abbreviations ACC: acetyl-coa carboxylase AICAR: 5-aminoimidazole-4-carboxamide-riboside AMPK: AMP-activated protein kinase ATPsynb: ATP synthase b-subunit BMI: body mass index CaM-K2: calmodulin-dependent protein kinase-2 CK-B: creatine kinase B subunit CPT1: carnitine palmitoyltransferase-1 CK1: casein kinase-1 CK2: casein kinase-2 DAG: diacylglycerol DMEM: Dulbecco's modified Eagle's medium EGF: epidermal growth factor EM: electron microscopic e-nos: endothelial nitric oxide synthase FDR T2DM : first degree relatives of patients with type 2 diabetes FFA: free fatty acids GDR: rates of glucose disposal GFA: glutamine:fructose-6-phosphate amidotransferase GIP: gastric-inhibotory-polypeptide GLP-1: glucagon-like-peptide-1 GRP78: 78-kDa glucose-regulated protein GS: glycogen synthase GSK-3: glycogen synthase kinase-3 G6P: glucose-6-phosphate HGP: hepatic glucose production HSP: heat shock protein IEF: isoelectric focusing IGT: impaired glucose tolerance IL-6: interleukin-6 IRS: insulin receptor substrate IRTK: insulin receptor tyrosine kinase 5

6 JNK: jun NH 2 -terminal kinase LCACoA: long-chain fatty acyl-coas NMR: nuclear magnetic resonance spectroscopy MRLC2: myosin regulatory light chain 2 MALDI-TOF-MS: matrix assisted laser desorption/ionization-time of flight mass spectrometry MAPK: mitogen-activated protein kinase MAPKAP-K2: MAPK-activated protein kinase 2 MS: mass spectrometry MW: molecular weight NEPHGE: non-equilibrium-ph-gradient electrophoresis NF-kB: nuclear factor kb NHS: normal human serum O-GLcNAc: O-linked N-acetylglucosamine OGT: UDP-GlcNAc:peptide glycosyl transferase PDK1: phospho-inositide dependent protein kinase-1 PGM-1: phosphoglucomutase-1 PhK: phosphorylase kinase PI3K: phosphatidylinositide 3-kinase PKA: protein kinase A PKB: protein kinase B or Akt PKC: protein kinase C PP1G.G M : muscle-specific glycogen-associated protein phosphatase 1 PP2A: protein phosphatase 2A ROS: reactive oxygen species TCA: trichloroacetic acid TNF-a: tumor necrosis factor-a T2DM: Type 2 diabetes mellitus UDP-GlcNAc: UDP-N-acetylglucosamine a1(vi)collagen: a1 chain of type VI collagen 6

7 TABLE OF CONTENTS Introduction Skeletal muscle insulin resistance: a hallmark feature of type 2 diabetes Insulin signaling into glycogen synthase in skeletal muscle Regulation of glycogen synthase activity in skeletal muscle Modulators of intracellular actions if insulin in skeletal muscle Alterations in insulin signaling in human skeletal muscle in type 2 diabetes Quantitative proteome analysis - a novel tool for the study of type 2 diabetes Aims of the thesis Metabolic and intracellular actions of insulin in skeletal muscle in type 2 diabetes Protein markers of type 2 diabetes in skeletal muscle in the fasting state Methods Assessment of metabolic and intracellular actions of insulin in skeletal muscle Quantitative proteome analysis of skeletal muscle Results and discussion Effect of insulin on glucose and lipid metabolism in type 2 diabetes Insulin signaling into skeletal muscle glycogen synthase in type 2 diabetes Phosphorylation of skeletal muscle glycogen synthase in type 2 diabetes Insulin action on metabolism, and signaling into and phosphorylation of glycogen synthase AMPK protein expression and activity in skeletal muscle in type 2 diabetes Effect of insulin on PP2A protein levels in skeletal muscle in type 2 diabetes Global protein expression profiling of skeletal muscle by proteome analysis Protein markers of type 2 diabetes in skeletal muscle in the fasting state Conclusions Hypothesis and future perspectives Summary Dansk resumé References Appendix: Papers I-IV 7

8 INTRODUCTION Skeletal muscle insulin resistance; a hallmark feature of type 2 diabetes Type 2 diabetes mellitus (T2DM) is the most common chronic metabolic disease in the world (1,2). A total of ~150 million people worldwide have been estimated to suffer from this heterogenous and complex disorder (1). T2DM is associated with increased morbidity and mortality, in particular due to cardiovascular complications, and results in a number of invalidating complications, which will affect the quality of life in an increasing number of subjects in the Western world (2,3). A better understanding of the pathophysiology of T2DM is therefore necessary in order to improve therapeutic and prevention strategies. Genes and environmental factors in T2DM. Despite no clear-cut Mendelian inheritance, there is plenty of evidence to suggest a strong genetic susceptibility for T2DM. In addition, there seems to be a predominantly maternal transmission of the disease in Caucasian populations (4-5). The life-time risk of developing T2DM has been estimated to be about 40% if one parent has T2DM (6), and up to 80% if both parents suffer from this disease (7-8). Furthermore, in previous studies the concordance rate of T2DM in identical twins was reported to approach 90% (9-10). However, in a recent twin study the concordance rate of T2DM in monozygotic twins (50%) was not significantly higher than in dizygotic twins (37%) implying a greater role for environmental factors than previously recognized (11). These factors include age, obesity and a sedentary lifestyle (12). Consistently, large prospective studies have shown that low fat diet, weight loss and increased physical activity significantly decrease the incidence of T2DM in subjects at risk for this disease (13-14). In addition, low birth weight has been demonstrated to be associated with later development of T2DM (15-16), and this seems to be caused by intrauterine malnutrition rather than a genetic predisposition (16-17). Thus, T2DM seems to be the result of a complex interplay between polygenetic predisposition and both pre- and postnatal environmental factors. Major abnormalities in glucose metabolism in T2DM. T2DM is pathophysiologically characterized by three major abnormalities of glucose metabolism; peripheral insulin resistance, defective insulin secretion and inappropriately elevated basal hepatic glucose production (3,18-19). The question as to which of these metabolic defects are primary or secondary has led to numerous studies of subjects at increased risk of developing T2DM. All three defects have been shown to precede manifest T2DM. Thus, several studies have demonstrated peripheral insulin resistance in glucose-tolerant first-degree relatives of patients with T2DM (FDR T2DM ), and there seems to be no doubt that this defect is a very early marker for the development of T2DM (20-24). Somewhat more controversial is whether defective insulin secretion is present before the prediabetic state defined 8

9 by impaired glucose tolerance (IGT) (8,19-21,24). However, at least some studies have reported both quantitative and qualitative defects in beta cell function including impaired first-phase secretion and impaired pulsatile release of insulin in FDR T2DM even before IGT (25-27). Increased basal hepatic glucose production (HGP) seems not to be present prior to the stage of IGT, when measured with appropriate methods (20,21,24,25). Nevertheless, the normal basal HGP despite elevated levels of fasting insulin in FDR T2DM may indicate hepatic insulin resistance at an earlier stage than IGT. In large prospective studies of normoglycemic FDR T2DM only peripheral insulin resistance has been shown to be a significant predictor of T2DM (8,23). It is therefore reasonable to assume that peripheral insulin resistance may be a primary abnormality of potential genetic origin that initially causes increased insulin production and secretion, which over time becomes inadequate and therefore causes impaired glucose tolerance. As the resulting postprandrial hyperglycemia contributes further to insulin resistance and defective insulin secretion - due to mechanisms, which is beyond the scope of this thesis to explain - a vicious cycle ensues that finally leads to frank T2DM. Defects in insulin action on muscle metabolism in T2DM. During infusion of insulin in physiological concentrations skeletal muscle is the major site of glucose disposal accounting for ~ 80% of whole body glucose clearance, in vivo (28-30). Furthermore, the use of nuclear magnetic resonance techniques (NMR) in vivo have shown that 80-90% of the glucose taken up in skeletal muscle during insulin stimulation is stored as glycogen (30). Accordingly, insulin resistance in skeletal muscle, defined as impaired glucose disposal during an euglycemic-hyperinsulinemic clamp, is a hallmark feature of T2DM, and is largely accounted for by reduced non-oxidative glucose metabolism (5,21,22,31-33), i.e decreased rates of glucose storage as glycogen (30). Therefore, at least quantitatively, impaired glycogen synthesis in muscle is considered to be the major defect of insulin-stimulated glucose metabolism in T2DM. However, other abnormalities in insulin action on skeletal muscle metabolism are consistently observed in patients with T2DM. Thus, insulin stimulation of glucose oxidation and suppression of lipid oxidation is also impaired, and is believed to contribute to insulin resistance in patients with T2DM (21,31,32,34-38). About 80-90% of patients with T2DM are obese, and as obesity itself causes skeletal muscle insulin resistance with the same defects in insulin action on glycogen synthesis, glucose and lipid oxidation, although to a lower extent (25,38-41), it is a matter of debate as to which of these abnormalities represent primary and perhaps genetically determined defects and which are secondary to changes in the metabolic milieu caused by obesity and T2DM. Most convincingly, impaired insulinstimulated glycogen synthesis has been reported in almost all studies of non-obese, glucose-tolerant 9

10 FDR T2DM (20-24,42), and in skeletal muscle cell cultures established from patients with T2DM (43), which strongly indicates a primary defect of possible genetic origin for this component. In obese FDR T2DM with IGT insulin-stimulated glucose oxidation was impaired compared to obese control subjects (44-45), which indicates that a component of this defect is not caused by obesity alone. Although, a few studies have reported impaired insulin-stimulated glucose oxidation in normoglycemic FDR T2DM (24,42), most studies of lean, glucose tolerant FDR T2DM could not find this abnormality (20-23,25). Therefore, a major part of the defect observed in obese FDR T2DM with IGT likely represent changes associated with this prediabetic state itself such as postprandrial hyperglycemia. Based on current available data it is, however, impossible to exclude that these abnormalities co-exist very early in the development of T2DM. In theory, insulin action on glucose and lipid oxidation may initially be spared at the expense of glycogen synthesis, which would explain the inability to detect altered glucose and lipid oxidation in lean, glucose tolerant FDR T2DM. Abnormalities in the fasting state in skeletal muscle metabolism in T2DM. In the past, primarily defects in the insulin-stimulated state have been studied. This may be due to the fact, that it is difficulty to assess substrate metabolism in skeletal muscle in the basal, fasting state. Indeed, basal whole-body glucose disposal, glucose oxidation, lipid oxidation and glucose storage can be estimated using the euglycemic-hyperinsulinemic clamp combined with [3-H 3 ]-glucose infusion and systemic indirect calorimetry. However, the glucose taken up by muscle in the basal, fasting state accounts for less than 20% of whole-body glucose disposal (28,34). To circumvent these methodological problems, limb-balance techniques have been developed allowing assessment of muscle metabolism across a large bed of muscle (34,46,47). From such experiments it is clear that abnormalities in glucose and lipid oxidation in skeletal muscle of type 2 diabetic subjects exist in the basal, fasting condition as well. Thus, conversely to what is observed in the insulin-stimulated state, glucose oxidation is increased and lipid oxidation impaired in the basal, fasting state in muscle of obese and type 2 diabetic subjects (34,38,40). This controversy between the basal, fasting and the insulin-stimulated states in insulin-resistant subjects has led to the hypothesis of metabolic inflexibility suggesting an impaired capacity to schwitch between carbohydrate and fat as oxidative energy sources as a major determinant of skeletal muscle insulin resistance (38). Although, these abnormalities in the basal, fasting state may largely represent adaptations to obesity and hyperglycemia, no data from experiments on FDR T2DM using the leg-balance technique are available. Therefore metabolic inflexibility as an integrated part of skeletal muscle insulin resistance in lean, glucose tolerant FDR T2DM cannot be excluded. In support of this hypothesis, artificial disruption of oxidative phosphorylation in mitochondria causes significant inhibition of both insulin-stimulated 10

11 glycogen synthesis and glucose uptake in muscle cells lines and animal muscle in vitro (48-49). More indirect support comes from studies suggesting that a reduced reliance on lipid oxidation in muscle during the basal, fasting state is likely a key mechanism by which excess amounts of triglyceride accumulate within skeletal muscle (38,40,50). Increased lipid content is in fact another key marker of skeletal muscle insulin resistance, which has been reported not only in obese and type 2 diabetic subjects (51-54), but also in lean, glucose tolerant FDR T2DM (55). Moreover, a close relationship between intramyocellular lipid concentrations (triglycerides) and insulin resistance have been reported in both healthy subjects, FDR T2DM, and type 2 diabetic subjects (50,51,53). Mitochondrial dysfunction or altered fiber type composition. A common denominator for the abnormalities observed in glucose and lipid oxidation in both the basal, fasting state and the insulin-stimulated state in muscle of insulin resistant subjects could be pertubations in skeletal muscle mitochondrial metabolism (38). Accordingly, some studies have reported reduced oxidative enzyme capacity, and reduced volumen, altered morphology and dysfunction of skeletal muscle mitochondria in T2DM, and to a lower extent in obesity (41,52,56-58). Again, a genetic component for mitochondrial dysfunction cannot be excluded, because similar studies of mitochondria in muscle of lean, glucose tolerant FDR T2DM have not been carried out. One possible explanation for the reduced oxidative enzyme activity, mitochondrial dysfunction and increased lipid content observed in obesity and T2DM is an altered proportion of muscle fiber types. Thus, in several studies an increased proportion of glycolytic, type 2 muscle fibers, and correspondingly a reduced proportion of oxidative, type 1 muscle fibers has been reported in obesity and T2DM (59-61). In a single study, increased amounts of glycolytic, type 2 muscle fibers were demonstrated in lean, glucose tolerant FDR T2DM (62), which indicates that an altered fiber type composition may represent another primary abnormality associated with skeletal muscle insulin resistance. On the other hand, in a study based on single-fiber analysis of the vastus lateralis muscle reduced oxidative enzyme activity and increased lipid content were present within each fiber type in obese and type 2 diabetic subjects (52). These data suggest that the contribution of mitochondrial dysfunction and increased lipid content to skeletal muscle insulin resistance is likely to be mediated by an effect of obesity and in particular T2DM on each fiber type. Other abnormalities associated with skeletal muscle insulin resistance. Within the last decade a number of other abnormalities than those introduced here have been reported to be associated with skeletal muscle insulin resistance in T2DM and FDR T2DM. This includes 1) abnormal release of hormones from the adipose tissue (adipocytokines) such as leptin, resistin, tumor necrosis factor-a (TNF-a), and adiponectin (63), 2) increased serum levels of inflammatory markers such 11

12 as C-reactive protein, fibrinogen and interleukin-6 (IL-6) indicating that a chronic low-grade systemic inflammation is involved in the pathogenesis of T2DM (64), and 3) defective secretion and action of gut incretin hormones such as glucagon-like-peptide-1 (GLP-1), gastric-inhibotorypolypeptide (GIP) and ghrelin (65). Although, indeed interesting, a more detailed description of these factors with respect to skeletal muscle insulin resistance in T2DM is felt to be beyond the scope of this thesis. Molecular mechanisms of metabolic abnormalities. In summary, metabolic studies suggest a number of metabolic abnormalities in skeletal muscle both in the basal, fasting and in the insulinstimulated state, which may represent very early markers for the development of T2DM. Of these, the most obvious candidate for a primary and perhaps genetically determined defect seems to be impaired insulin stimulation of glycogen synthesis. Although metabolic studies are fundamental to understand the pathophysiology of skeletal muscle insulin resistance in T2DM, such studies will not reveal the underlying biochemical and molecular mechanisms. It is therefore necessary to extend the field of research to include studies of the proteins and enzymes accounting for the abnormalities described above. In order to find an explanation for the impaired insulin activation of glycogen synthesis, studies are often based on a prior knowledge of the possible mechanisms by which insulin mediates its effects at a cellular level, here termed the classical approach. In contrast, when it comes to the molecular mechanisms responsible for metabolic inflexibility, only a few enzymes, so-called metabolic fuel regulators or nutrient sensors, have been identified until now. Therefore, the application of more global approaches should be considered, in order to identify abnormalities in the potential numerous proteins and enzymes assumed to be involved in the cross-talking between the systems handling the uptake, storage and oxidation of glucose and lipids. Insulin signaling into glycogen synthase in skeletal muscle In order to elucidate the pathophysiology of insulin resistance in T2DM, it is important to understand the mechanisms by which insulin signals to the cell interior, and in particular into glycogen synthase (GS). A very brief review of studies that have given valuable insight in to the intracellular actions of insulin is therefore presented below. Numereous intracelluar actions of insulin. Insulin is a potent anabolic hormone that modulates a wide variety of biological processes in skeletal muscle including glycogen synthesis, glucose transport, protein synthesis and mitogenesis (Fig 1) (66). The intracellular actions of insulin is mediated by modification of the activity and the subcellular location of key regulatory proteins and enzymes (kinases and phosphatases) primarily by affecting their phosphorylation state (67). 12

13 Insulin also controls the amount of numerous proteins through an action on translational and transcriptional factors giving rise to regulation of gene expression (68). Promotion of glycogen synthesis is one of the key biological actions of insulin in skeletal muscle, and as noted above, a defect in this action of insulin seems to be a major determinant of skeletal muscle insulin resistance in type 2 diabetic subjects and FDR T2DM. Therefore, the signaling mechanisms by which insulin stimulates glycogen synthesis have been the subject of numerous studies, including the present Ph.D.-study. Figure 1. Schematic illustration of major signaling pathways of insulin action. The phosphorylated insulin receptor binds and phosphorylates IRS proteins and Shc, which bind differentially to various downstream signaling proteins. PI3-kinase is critical for metabolic actions of insulin, such as glucose transport, glycogen synthesis, and protein synthesis, whereas Grb-2/ SOS complex, which activates the MAP kinase cascade, is critical in mitogenic response. PI3-kinase probably modulates the mitogenic response as well. Reproduced from ref. 66. The proximal steps in insulin signaling. Insulin signaling involves a cascade of events initiated by insulin binding to its cell-surface receptor. This causes autophosphorylation and subsequent binding and activation of the insulin receptor substrate (IRS) proteins, which leads to activation of the phosphatidylinositide (PI) 3-kinase (3K) and subsequent formation of PI(3,4,5)P 3 from PI(3,4,5)P 2 in the inner part of the plasma membrane (66,67,69,70). These initial, proximal steps are necessary for the initiation of a number of divergent signaling cascades mediating nearly all the metabolic actions of insulin, e.g. on glucose transport, glycogen synthesis and protein synthesis in skeletal muscle (Fig. 1+2). A pathway for activation of a GS phosphatase. Identification of the distal signaling elements involved in insulin action on glycogen synthesis has been a major challenge for the last three decades. Glycogen synthase (GS) catalyzes the last step in the pathway of glycogen synthesis, and as early as in 1963 it was demonstrated that insulin activated GS by promoting its dephosphorylation (67). As discussed later, this can in theory be achieved either by activation of 13

14 a GS phosphatase or by inhibition of a GS kinase. Of the many potential kinases and phosphatases affecting GS phosphorylation in vitro, only one, before 1992, seemed to be significantly regulated by insulin (67,69). Thus, it was shown that the muscle-specific glycogen-associated type 1 protein phosphatase (PP1G.G M ) was stimulated by insulin (71), and this phosphatase was believed to be responsible for direct activation of GS by dephosphorylation of all the sites in GS (72-73). Subsequently, a kinase responsible for the activation of PP1G.G M was identified as the RSK2 isoform of p90rsk (71,74). This enzyme was known to be activated by the mitogen-activated protein kinase (MAPK) (75). These observations led to the proposal of a model in 1993 in which activation of PP1G.G M by insulin was mediated by the MAPK signaling pathway (Fig. 2) (72). However, several lines of evidence indicated that the MAPK pathway could not be involved in the activation of GS by insulin (69). Thus, use of another external stimuli, epidermal growth factor (EGF), known to activate the MAPK pathway, did not activate GS (76), and the use of inhibitors at different steps in this pathway, showed that inhibition of the pathway distal to PI3K did not inhibit insulin activation of GS (76-77). These studies led to the conclusion that the activation of MAPK and RSK2 was neither necessary nor sufficient for the activation of GS in skeletal muscle, but did not preclude the possibility that insulin activates GS through activation of PP1G.G M (69). Isoforms of p90rsk other than RSK2 can activate PP1G.G M, and thereby explain PP1G.G M mediated activation of GS (78). Thus, it was reported that the RSK3 isoform of p90rsk could be activated by the Jun NH 2 -terminal kinase (JNK) in vitro, and that activation of JNK by insulin in vivo preceded the activation of PP1G.G M and GS (79). It was therefore stated that JNK was involved in insulin activation of GS (79). However, there is no direct evidence to implicate JNK in the activation of GS, and the transient (4 min) activation of JNK in contrast to the stable effect of insulin on GS (hours) (69), and a much stronger effect of EGF than insulin on RSK3 activity (80), indicated that these enzymes were not distal components of the insulin signaling cascade leading to activation of GS. In fact, elevated activity of JNK induced by obesity and TNF-a seems to impair proximal insulin signaling by inhibitory phosphorylation of IRS-1 at Ser 307 (81). A pathway for inhibition of a GS kinase. Insulin mediated dephosphorylation (activation) of GS needs not involve activation of a phosphatase. Thus, the phosphate turnover in GS is relatively rapid, and net dephosphorylation could result from the inhibition of appropriate kinases (82). Since 1992 an increasing number of studies including in human skeletal muscle in vivo have demonstrated that insulin decreases the activity of glycogen synthase kinase-3 (GSK-3) (83-86), and this inhibition seems to be sufficient to account for the observed activation of GS by insulin (67). The inhibition of GSK-3 induced by insulin was shown to be mediated through phosphorylation of 14

15 Ser21 in GSK-3a and Ser9 in GSK-3b (85). Using different inhibitors of PI3K it was demonstrated that insulin-mediated inhibition of GSK-3 was catalyzed by a kinase whose activation by insulin was dependent on PI3K activity (85). From a series of studies a protein kinase was identified, termed protein kinase B (PKB) or Akt, which was activated by insulin and which phosphorylated and inactivated GSK-3 (85-86). Akt was shown to be activated by phosphorylation of two sites at Thr308 and Ser473 (87), and later an enzyme capable of phosphorylating Akt at Thr308 (but not Ser473) was identified (88). This enzyme was only active in the presence of PI(3,4,5)P 3 or PI(3,4)P 2, and was therefore termed phospho-inositide dependent protein kinase-1 (PDK1) (88). The identity of the kinase responsible for the phopshorylation of Ser473 (putatively called PDK2) remains elusive (67,70). In this way the gap between the distal and proximal ends in insulin signaling into GS was closed, and it is currently believed that insulin signaling from the insulin receptor into GS involves activation of IRS-1 and -2, PI3K, PDK1 (and PDK2), and Akt, which in turn leads to inhibition of GSK-3, and hence dephosphorylation and activation of GS (Fig. 2). Fig. 2. Signaling pathways formerly (dark symbols) and currently (light symbols) believed to mediate the activation of glycogen synthase (GS) by insulin, and the pathway of glycogen synthesis in skeletal muscle. Adapted from ref. 69. See abbreviation list and ref. 69 for abbreviations. A pathway for GLUT4 translocation. Another important metabolic response to insulin in skeletal muscle is the stimulation of glucose transport, which is largely due to the translocation of the insulin-regulated glucose transporter, GLUT4, from its intracellular storage pool to the plasma membrane (89). As mentioned above, activation of PI3K is involved in the effect of insulin on glucose transport (70,90), and there is now accumulating evidence that the insulin signaling process 15

16 that promotes GLUT4 translocation to the plasma membrane involves activation of Akt (Fig. 2) (70), although evidence to the contrary is also available (91). There is data to suggest that full stimulation of glucose transport by insulin requires the activation of another kinase, the z isoenzyme of protein kinase C (PKC), and that this kinase may work by potentiating the effect of Akt activation on glucose transport (70). However, many questions concerning the molecular mechanism by which insulin modulates glucose transport into skeletal muscle remain unanswered. Rate-limiting step of glycogen synthesis. Despite progress in the unraveling of the insulin signal transduction pathways mediating insulin activation of GS and glucose transport, debate continued whether insulin stimulation of glycogen synthesis involves a push or a pull - or in other words, what is the rate-limiting step for glycogen synthesis in skeletal muscle? According to the push hypothesis insulin primarily acts by activating glucose transport, thereby pushing glucose into glycogen. In the pull hypothesis glucose and its metabolites are pulled into glycogen by the activation of GS (3,69) (Fig. 2). Support for the push hypothesis came from studies of transgenic mice overexpressing GLUT1 in skeletal muscle. In these mice both glucose transport and glycogen levels were increased several-fold despite no increase in GS activity compared to non-transgenic litter mates (92-93). In addition, a model proposed by Shulman and collegues, which is based on NMR studies of intracellular metabolites during insulin stimulation, and which is in essence an extension of the push hypothesis, suggests that most control of glycogen synthesis is exerted at the stage of glucose transport/phosphorylation, and that insulin activation of GS has merely evolved to prevent overaccumulation of metabolites such as glucose-6- phosphate (G6P) (94). However, in transgenic mice overexpressing GLUT4, the major insulinresponsive glucose transporter in skeletal muscle, insulin-stimulated glucose transport was increased without any change in the rate of glycogen synthesis or glycogen levels (95-96), which argues against the push hypothesis as the only contributor to glycogen synthesis. In further support of the pull hypothesis studies of skeletal muscle from transgenic mice overexpressing a constitutively activated form of GS demonstrated five-fold higher glycogen levels without any change in GLUT4 content or basal or insulin-stimulated glucose uptake compared to transgenic litter mates (97). If glucose transport was rate-limiting for glycogen synthesis, increased GS activity without a change in glucose transport should not have increased glycogen levels, and these data therefore support an active role of GS in this process (69). Based on these studies it is generally accepted that both glucose transport and glycogen synthase contributes to the control of glycogen synthesis in skeletal muscle (98). 16

17 Regulation of glycogen synthase activity in skeletal muscle As mentioned earlier, stimulation of glycogen synthesis is one of the key biological actions of insulin in skeletal muscle, and a defect in this action seems to play a major role for insulin resistance in T2DM. Glycogen synthase, the enzyme which catalyzes the last step in the pathway of glycogen synthesis, was in fact the first enzyme whose activity was shown to be regulated by insulin. Thus, for four decades ago it was shown that GS was inactivated by phosphorylation and reactivated by dephosphorylation, and that insulin stimulated GS by promoting its dephosphorylation (67,69). Since that time numerous studies have added to reveal the complexity of the regulation of GS. Multisite phosphorylation of GS. In brief, these studies have revealed that skeletal muscle GS activity is controlled by multisite phosphorylation and several allosteric effectors, of which G6P seems to be the most important (67,69,99-102). Phosphorylation leads to inactivation of GS, but full activity can be restored in the presence of G6P (67,69,99). Of the nine serine residues, which are phosphorylated in mammalian GS in vivo (73,103) the sites most important for activating the enzyme are sites 2 and 2a in the NH 2 -terminus and sites 3a and 3b in the COOH-terminus (73, ), and the effects seem to be additive so that almost complete inhibition of GS occurs if all four sites are phosphorylated ( ) (Fig.3). In contrast, phosphorylation of GS at sites 1a, 1b, 4 and 5 have very little, if any, effect on the enzyme activity (69,103,105), but at least in theory these sites may play a role in the localization of GS and/or the regulation of allosteric effectors. GS kinases and phosphatases. Identification of the kinases and phosphatases that regulate phosphorylation of GS at these multiple sites was another major challenge. Thus, it was realized that many of the kinases and phosphatases identified in vitro may have no physiological relevance in vivo (67,69). Nine kinases and one phosphatase are believed to be potential regulators of GS activity (Fig. 3). Of these, GSK-3 is the most active kinase phosphorylating sites 4, 3c, 3b and 3a in vitro, and seems to do so in a sequentiel manner (67,69, ). Moreover, these four sites are only phosphorylated by GSK-3 if site 5 has already been phosphorylated by casein kinase-2 (CK2), a phenomenon termed hierarchal phosphorylation ( ). Interestingly, CK2 is believed to be constitutively active in vivo. Protein kinase A (PKA), which is known to be activated by increases in the second messenger camp, preferentially phosphorylates sites 2, 1a and 1b in vitro (101). Site 2 can be phosphorylated by several other protein kinases in vitro, of which PKC, AMP-activated protein kinase (AMPK), MAPK-activated protein-k2 (MAPKAP-K2), phosphorylase kinase (PhK), and calmodulin-dependent protein kinase-2 (CaM-K2) may have physiological relevance under different conditions (67, ). Casein kinase-1 (CK1) is the only detectable site 2a kinase in 17

18 skeletal muscle extracts (109). Prior phosphorylation of site 2 has been shown to increase several fold the phosphorylation of site 2a by CK1 (109,113), demonstrating another example of hierarchal phosphorylation in the regulation of GS (100). A number of phosphatases, including protein phosphatase 2A (PP2A), can dephosphorylate GS in vitro (114). However, in muscle PP1 has been reported to be the dominating protein phosphatase that dephosphorylates and activates GS ( ). Moreover, PP1 is the only phosphatase that specifically associates with glycogen particles, and is thus localized in the compartment of muscle where it is expected to work (115). The glycogen associated form from muscle, termed PP1G.G M is thought to be responsible for activation of GS by direct dephosphorylation (71-73). In vitro, PP1G.G M can dephosphorylate all of the sites in GS (71-73). Fig. 3. Phosphorylation sites in the NH 2 - and COOH terminal ends of glycogen synthase (GS), and the kinases and phosphatase believed to play a role for the regulation of GS. The most important sites are marked with black. PP1G.G M can dephosphorylate all of the sites in GS. Horizontal arrow denotes hierarchal phosphorylation. See abbreviation list and ref. 69 for abbreviations. Effect of insulin on GS phosphorylation. In vivo, studies on rabbit skeletal muscle have shown that most of the phosphate released from GS in response to insulin was removed from the tryptic peptide containing sites 3a and 3b (73). However, studies in cells expressing rabbit skeletal muscle GS and in rat skeletal muscle in vitro have demonstrated that insulin in high concentrations promote dephosphorylation of sites 2 and 2a as well (104,105,107,108). Insulin-mediated stimulation of PP1G.G M was originally thought to be responsible for the activation of GS by promoting dephosphorylation of both NH 2 - and COOH-terminal sites on GS (71-72). Although, 18

19 some recent reports have questioned the ability of insulin to activate GS by PP1G.G M (116,117), as well as the mechanism by which PP1G.G M is activated by insulin (118), an even more recent study indicates that insulin does in fact activate PP1G.G M (119). Insulin has in several studies been shown to inhibit GSK-3 in skeletal muscle to a degree almost sufficient to account for the observed activation of GS (83-86). However, the final step, that GSK-3 inhibition leads to dephosphorylation of sites 3a and 3b and subsequent activation of GS in vivo has yet to be demonstrated. Moreover, recent studies have indicated that sites 3a and 3b can be phosphorylated by other currently unknown protein kinases (69,105,106), and that GSK-3 is not essential for GS activation by insulin, at least not in Rat-1 fibroblasts expressing rabbit skeletal muscle GS(104). Despite, some lack of clarity in this field, GSK-3 and PP1G.G M are still believed to be the best candidates for identification of the insulin signaling pathway(s) responsible for activation of glycogen synthesis. However, with respect to the skeletal muscle insulin resistance, increase in any kinase capable of phosphorylating GS may be of pathophysiological relevance. Thus, increased phosphorylation of GS at sites not regulated by insulin in vivo may counteract the effect of insulin on other sites, and thus impair glycogen synthesis and glucose disposal. Assessment of GS activity and phosphorylation. Another important issue, is the method by which GS-activity is measured. Thus, in studies of human skeletal muscle biopsies, GS-activity is measured in vitro as the amount of the substrate UDP-glucose incorporated into glycogen in the presence of no, low or high concentrations of G6P (120,121). Based on such measurements rough estimates of the in vivo activity of GS can be calculated, which is believed to represent a measure of the phosphorylation state of GS, and to be independent of the actual muscle G6P content. For these assumptions to be valid, GS activity must not be regulated covalently by any other means than phosphorylation, and the actual muscle G6P content must have no influence on the phosphorylation state of GS. In fact, there are some data to suggest that G6P controls the phosphorylation state of GS in muscle by inhibiting a glycogen synthase kinase, i.e. PKA (99). Such a mechanism would explain the finding of normal insulin-stimulated GS-activity in skeletal muscle of patients with T2DM, when studied during hyperglycemic clamp conditions in which G6P content in skeletal muscle biopsies seems to be increased (122,123). Furthermore, in a recent study it was demonstrated that muscle GS is modified by O-linked N-acetylglucosamine (O-GlcNAc), and data indicated that O-glycosylation may inhibit GS activity without increasing phosphorylation (124). Glycosylation seems to increase with long term exposure to high glucose concentrations (124), and may thus in part explain impaired insulin activation of muscle GS activity in both type 1 and type 2 diabetic subjects, even without affecting insulin-mediated dephosphorylation of GS. Therefore, in order to 19

20 obtain valid information about the effect of insulin on the phosphorylation state of GS, novel methods able to precisely characterize the specific phosphorylation sites involved are warranted. This will also help to identify the kinases and phosphatases of physiological relevance for phosphorylation of GS under these conditions. Although increased electrophoretic migration of GS immunoreactivity on SDS-PAGE gel from an apparent molecular weight (MW) of 84 kda to kda during hyperinsulinemia may indeed reflect the dephosphorylating effect of insulin on GS (125), data concerning the contribution of the different phosphorylation sites to enzyme activity in human muscle GS have so far not been available. Modulators of intracellular actions of insulin in skeletal muscle. With the rapidly progress in the understanding of the molecular mechanism underlying the metabolic actions of insulin, numerous possible modulators of the enzymes involved in insulin action have emerged. These modulators may be key regulators of nutrient sensing, stress signaling and mitochondrial function. The AMPK system. AMPK is a signal intermediate in metabolic regulation in mammalian cells, and may serve as a metabolic master switch in response to alterations in cellular energy charge such as ATP/AMP and creatine phosphate/creatine ratios ( ). In skeletal muscle AMPK has been implicated in the regulation of lipid oxidation, glucose transport and GS activity (112, ). In brief, acute activation of AMPK leads to phosphorylation and inhibition of acetyl-coa carboxylase (ACC), and hence a decline in malonyl-coa, which in turn relieves the inhibition of carnitine palmitoyltransferase-1 (CPT1), the enzyme that controls the transfer of long-chain fatty acyl-coas (LCACoAs) into mitochondria where they are oxidized (126). Reports of an inhibitory effect of insulin on AMPK activity in animal cardiac and skeletal muscle ( ), suggest that inhibition of AMPK may contribute to insulin-mediated activation of GS and suppression of lipid oxidation. Consistently, both biochemical and physiological studies in vitro suggest that AMPK inhibits GS activity by phosphorylation at site 2 (Fig. 3) (112,130, ). Decreased AMPK activity in the basal, fasting state will result in impaired lipid oxidation, and hence accumulation of intramyocellular lipids. On the other hand, increased AMPK activity may impair insulin-mediated suppression of lipid oxidation. The AMPK is a kinase fully activated as a heterotrimeric complex consisting of a catalytic (a) and two regulatory (b, g) subunits ( ). Two isoforms of the catalytic subunit (a 1, a 2), two of the b subunit, and three of the g subunit, have been identified in mammalian cells ( ). The a1 subunit is widely distributed, whereas a2 is primarily found in skeletal muscle, heart and liver (140). Apparently b2 is expressed to a greater extent in skeletal 20

21 muscle than is b1 (138). g1 and g2 mrna are found in a variety of tissues, whereas g3 mrna was only detected in skeletal muscle, however the protein appears to be more broadly distributed (139). There are increasing evidence to suggest a role for the AMPK system in insulin sensitivity, glucose uptake and glycogen synthesis. Thus, AMPK subunit expression is changed in response to exercise training ( ). This may be involved in the up-regulation of gene-expression in response to exercise training and/or in the metabolic alterations seen after exercise training, e.g. enhanced peripheral insulin sensitivity. In addition, the RN - mutation, located in the PRKAG3 gene encoding the AMPK g3 subunit, causes substantial increase in glycogen content in pig muscle (144).Whether this is due to a decreased inhibitory effect of AMPK on GS or a sustained activation of glucose uptake is currently debated, but studies of a similar mutation in the g1 subunit are in favor of an activating function (128,145). The whole body a2-deficient mouse is insulin resistant possibly due to enhanced sympathetic nervous activity (146). Furthermore, acute AMPK activation seems to be involved, at least partially, in contraction-induced glucose uptake and GLUT4 translocation, although the intracellular mechanisms are not clear at present (147). Chronic activation of AMPK is associated with increased mitochondrial biogenesis, improved insulin sensitivity and increases in GLUT4 protein content as well as in the activity of mitochondrial oxidative enzymes, thus, mimicking exercise training ( ). For these reasons, alterations in the AMPK signaling system have been hypothesized to play a role in the pathogenesis of T2DM (126). PKC and PP2A in lipid metabolism. As noted above, insulin resistance is associated with obesity, increased circulating free fatty acid (FFA) levels and intramyocellular lipid content in muscle (156). Studies of insulin resistance in T2DM by lipid infusion, obesity, fat-fed animals and lipid-treated cells have shown that this might involve interference with insulin signaling transduction and GS activity by lipid-activated signaling pathways. FFAs are activated to LCACoAs Fig.4. Modulation of insulin signaling transduction into glycogen synthase by lipid-activated signaling pathways. FABP, fatty acid binding protein; FAT/CD36, fatty acid transporter. 21

22 before transport across mitochondrial membrane by CPT1, and LCACoAs are also the substrates for esterification of FFAs for the synthesis of triglycerides. Thus, excess triglyceride and FFA in muscle in insulin resistant states will result in increased LCACoAs levels, which in turn lead to increased diacylglycerol (DAG). DAG activates many isoforms of the PKC family. Increased activity of different PKC isoforms may inhibit proximal insulin signaling transduction by inhibitory serine/threonine phosphorylation of IR and IRS-1 ( ), but may also cause direct inactivation of GS by phosphorylation at site 2 (111) (Fig. 4). Increased lipid content in muscle also causes increased de novo synthesis of ceramide, which has been shown to inhibit distal sites of the insulin signaling pathway such as Akt and GSK-3 (157,161,162). This seems to be brought about by activation of PP2A (157, 162) (Fig. 4). Interestingly, insulin have been shown to inhibit PP2A in cultured muscle cells (114,162), and PP2A have been shown to inhibit Akt, and to activate GSK-3 (74, ). Unfortunately, PP2A is less well studied in human skeletal muscle insulin resistance. However, altered activity of certain PKC isoforms and PP2A are potential mediators of impaired insulin signaling. Fig. 5. Proposed general theory of how elevated FFA and hyperglycemia result in the pathophysiology of diabetes via the generation of ROS. This diagram shows the proposed causative link between hyperglycemia, elevated FFA, mitochondrial ROS generation, oxidative stress, activation of stress-sensitive pathways (p38 MAPK, JNK/SAPK, NF-kB and others), insulin resistance, b-cell dysfunction, and diabetic compli-cations. Adapted from ref For abbreviations see text. Stress signaling. As reviewed extensively recently, there is accumulating evidence to suggest that oxidative stress and activation of stress-activated signaling pathways contribute to insulin resistance in a variety of tissues (165). In brief, both elevated glucose and FFA levels cause oxidative stress due to increased production of mitochondrial reactive oxygen species (ROS) (165). Oxidative stress leads to the activation of protein kinases (p38 MAPK, JNK, nuclear factor-kb (NFkB)) in several stress-sensitive pathways, which seems to blunt the effects of insulin on glucose transport and glycogen synthesis (49,165) (Fig. 5). The mechanisms by which these stress kinases impair insulin signaling are unclear, but may involve serine phosphorylation of IRS proteins and 22